Is this microsoft stating that they aren't able to get acceptable reliability from Azure? (I mean, I think a lot of us have heard that, but it's interesting to hear it from microsoft themselves).
openai, anthropic, google and a plethora of chinese models all end up pushing code into github. you can discuss whether gpt 5.5 is better than opus 4.7, but for github it doesn't matter: they'll be receiving the code no matter which llm spits it out.
amazing on one hand, quite scary on the other for github and all other forges if this continues and there is no reason why it wouldn't.
Glad that they released some data about new repo/issues/commits over the last years. It confirms what everyone else already believed from the outside: agents are putting a lot of extra, sudden pressure on GitHub. It's like a startup that is growing exponentially, with the difference that they already have a large user base to serve - and that keeps them in the bullseye - and probably a not-so-fast-moving organization when it comes down to changes. On the other side of the coin, they also have a lot of talent, infra and money a startup might not have yet.
Stop subsidizing tokens now that we extracted enough training data from you and we have enough agentic junkies business to keep the flywheel going up and cut on the loss leaders. [0]
> While we were already in progress of migrating out of our smaller custom data centers into public cloud, we started working on path to multi cloud. This longer-term measure is necessary to achieve the level of resilience, low latency, and flexibility that will be needed in the future.
It's kind of hard to read this with a straight face.
The unlabelled graph with big numbers on top, the priorities that don't match with what we're experiencing, and a list of things that they're doing without a real acknowledgement of the _dire_ uptime over the last 12 months....
Personally, I’m sympathetic. We know that GitHub did a huge amount of work over the last decade to make Git scale, which has benefited us all. These new scaling challenges are real challenges, 30x growth would be a nightmare for any system that was already pushing the limits of what was possible, I think we are being far too hard on GitHub, they deserve a little grace.
I'm biased (founder of tangled.org), but the future really should be federated forges. Host repositories on sovereign infra with global identity + federated "metadata" (issues, pulls, etc.).
Global indices for this should be trivial to spin up so availability is never a concern (we're working towards this!).
* we had to resolve a variety of bottlenecks that appeared faster than expected from moving webhooks to a different backend (out of MySQL)
* * redesigning user session cache to redoing authentication and authorization flows to substantially reduce database load.
* we accelerated parts of migrating performance or scale sensitive code out of Ruby monolith into Go.
I'd like to know what database backend they migrated to. I was also surprised to read that the migration from Ruby to a more performant language had not already been completed. I assume this is because it a large code base with many moving parts, etc.
Hah, love that now they say "Our priorities are clear: availability first, then capacity, then new features" when 6 months ago, it was seemingly exactly the same except Azure supposedly was gonna save them:
> GitHub Will Prioritize Migrating to Azure Over Feature Development - GitHub is working on migrating all of its infrastructure to Azure, even though this means it'll have to delay some feature development.
> In a message to GitHub’s staff, CTO Vladimir Fedorov notes that GitHub is constrained on capacity in its Virginia data center. “It’s existential for us to keep up with the demands of AI and Copilot, which are changing how people use GitHub,” he writes.
So the currently delayed feature development is now gonna be further delayed, yet almost every week we see new features and changes, just the other day the single issues view was changed, as just one example. And it was "existential" 6 months ago yet they keep stumbling on the exact same issue today?
Even if they're focused exclusively on reliability and uptime, we get the experience that we have today, kind of incredible how a company with the resources of Microsoft seemingly are unable to stop continuously shot themselves in the foot. It's kind of impressive actually. As icing on the cake, they've decided to buy up all popular developer services then migrate them all to the same platform, great idea too.
I wonder if they’ll end the free lunch we’ve been having since the MS takeover. There’s been a deluge of spam and crapware projects due to the LLM wave which is visible in that graph. Can’t see them sustaining being a public dustbin for low value projects forever.
> Vladimir Fedorov is GitHub's Chief Technology Officer .... He currently serves on the board of Codepath.org, an organization dedicated to reprogramming higher education to create the first AI-native generation of engineers, CTOs, and founders.
> The main driver is a rapid change in how software is being built. Since the second half of December 2025, agentic development workflows have accelerated sharply.
GitHub instability has started way before that. I understand it’s too much to ask of a trillion-dollar corporation to consider the impact of their own actions, but perhaps they should’ve thought of that before forcing LLM development down everyone’s throats.
I know that I'm simplifying (probably too much), but it seems like things were fine when GitHub was still a Ruby on Rails monolith and all the rigmarole with microservices etc. only made things worse.
This latest incident was the nail in the coffin for me. I've been on GitHub since 2012 but I'm feeling the pull to migrate out to Gitea/Forgejo. Has anybody done this recently? How'd it go?
anyone who's actually worked there, could you explain why they're finding scalability and reliability so hard? naively it seems like 'repo groups', ie clusters of repositories linked by being mutual forks, would be fairly isolated for the whole git storage layer, and everything else feels pretty easily parallelisable (issues, actions, etc, modulo taking locks now and then to submit results or whatever). and given that, surely you can incrementally deploy changes across those many shards to avoid most big outages?
are there big conceptual serialisations that I've missed? is it just not well factored? was the move to Azure just a catastrophically bad idea? some other thing?
I can't speak to the last few years since I left, but over the many years I was there the git storage layer was almost never the core issue - it was well designed by infrastructure-minded nerds that leveraged and improved git and replicated it really well across multiple nodes.
What always struggled was the richness of the Rails monolith itself and its backing MySQL databases - the expectation that everything links to everything throughout the product (think: issue cross references across orgs that only appear if you're able to access the remote repo, and other things like that).
Those details appear richly everywhere you look, and the combination of that with a general lack of understanding and/or focus on performance (shipping big features is hard, shipping them with performance at scale is MUCH harder), compounded by Ruby being an easy language to get performance wrong in (object count really hurts, and it makes it very easy to create many) leaves every feature adding to the performance problem, and makes it daunting/impossible to make fast once it's slow.
There was a full on year or more of making GitHub fast while I was there that just couldn't gain enough momentum to make enough of a dent to make it better. I remember finding and fixing a N^3 (or maybe it was N^4? something bad) in the home page activity feed - the worst thing I found but gives an idea. IMO it would need a fresh view of how to keep interfaces simple and how to design the data layers performantly - not adding every bell and whistle to every screen.
I hope someone at GitHub realises they are about to lose everything that was hard earned by early GitHub - it once was a site people (myself included) looked up to for ideal availability, responsible releases, data driven improvement - but no more it seems :(
Not enough attention is being put in the production/delivery mismatch.
GitHub is claiming they require 30x scale due to the giant increase in repository creation, PRs, commits, etc.
I have not seen a single product increase in features or quality as an end user, nor new significant products have come out in this period (other than the LLMs themselves).
I can not figure out what on Earth they've done with these graphs, it almost seems like these are an artists impression of a graph.
Looking at the commit graph: Why do commits have big steps followed by slow rolloffs? Why do the steps not happen at uniform points Why do larger steps sometimes have less of a slope than smaller steps but not all the time?
Then looking at the other graphs there's completely different effects going on.
70 comments
[ 2.8 ms ] story [ 71.6 ms ] threadIs this microsoft stating that they aren't able to get acceptable reliability from Azure? (I mean, I think a lot of us have heard that, but it's interesting to hear it from microsoft themselves).
amazing on one hand, quite scary on the other for github and all other forges if this continues and there is no reason why it wouldn't.
Stop subsidizing tokens now that we extracted enough training data from you and we have enough agentic junkies business to keep the flywheel going up and cut on the loss leaders. [0]
[0] https://news.ycombinator.com/item?id=47923357
Wild
The unlabelled graph with big numbers on top, the priorities that don't match with what we're experiencing, and a list of things that they're doing without a real acknowledgement of the _dire_ uptime over the last 12 months....
Global indices for this should be trivial to spin up so availability is never a concern (we're working towards this!).
* we had to resolve a variety of bottlenecks that appeared faster than expected from moving webhooks to a different backend (out of MySQL)
* * redesigning user session cache to redoing authentication and authorization flows to substantially reduce database load.
* we accelerated parts of migrating performance or scale sensitive code out of Ruby monolith into Go.
I'd like to know what database backend they migrated to. I was also surprised to read that the migration from Ruby to a more performant language had not already been completed. I assume this is because it a large code base with many moving parts, etc.
That's a delayed April fool's right?
> GitHub Will Prioritize Migrating to Azure Over Feature Development - GitHub is working on migrating all of its infrastructure to Azure, even though this means it'll have to delay some feature development.
> In a message to GitHub’s staff, CTO Vladimir Fedorov notes that GitHub is constrained on capacity in its Virginia data center. “It’s existential for us to keep up with the demands of AI and Copilot, which are changing how people use GitHub,” he writes.
https://thenewstack.io/github-will-prioritize-migrating-to-a...
So the currently delayed feature development is now gonna be further delayed, yet almost every week we see new features and changes, just the other day the single issues view was changed, as just one example. And it was "existential" 6 months ago yet they keep stumbling on the exact same issue today?
Even if they're focused exclusively on reliability and uptime, we get the experience that we have today, kind of incredible how a company with the resources of Microsoft seemingly are unable to stop continuously shot themselves in the foot. It's kind of impressive actually. As icing on the cake, they've decided to buy up all popular developer services then migrate them all to the same platform, great idea too.
I think I found the issue.
GitHub instability has started way before that. I understand it’s too much to ask of a trillion-dollar corporation to consider the impact of their own actions, but perhaps they should’ve thought of that before forcing LLM development down everyone’s throats.
are there big conceptual serialisations that I've missed? is it just not well factored? was the move to Azure just a catastrophically bad idea? some other thing?
What always struggled was the richness of the Rails monolith itself and its backing MySQL databases - the expectation that everything links to everything throughout the product (think: issue cross references across orgs that only appear if you're able to access the remote repo, and other things like that).
Those details appear richly everywhere you look, and the combination of that with a general lack of understanding and/or focus on performance (shipping big features is hard, shipping them with performance at scale is MUCH harder), compounded by Ruby being an easy language to get performance wrong in (object count really hurts, and it makes it very easy to create many) leaves every feature adding to the performance problem, and makes it daunting/impossible to make fast once it's slow.
There was a full on year or more of making GitHub fast while I was there that just couldn't gain enough momentum to make enough of a dent to make it better. I remember finding and fixing a N^3 (or maybe it was N^4? something bad) in the home page activity feed - the worst thing I found but gives an idea. IMO it would need a fresh view of how to keep interfaces simple and how to design the data layers performantly - not adding every bell and whistle to every screen.
I hope someone at GitHub realises they are about to lose everything that was hard earned by early GitHub - it once was a site people (myself included) looked up to for ideal availability, responsible releases, data driven improvement - but no more it seems :(
GitHub is claiming they require 30x scale due to the giant increase in repository creation, PRs, commits, etc.
I have not seen a single product increase in features or quality as an end user, nor new significant products have come out in this period (other than the LLMs themselves).
Where is all this code going?
Looking at the commit graph: Why do commits have big steps followed by slow rolloffs? Why do the steps not happen at uniform points Why do larger steps sometimes have less of a slope than smaller steps but not all the time?
Then looking at the other graphs there's completely different effects going on.